TY - JOUR
T1 - Machine learning-based COVID-19 acute respiratory distress syndrome phenotyping and clinical outcomes
T2 - A systematic review
AU - Tenda, Eric Daniel
AU - Henrina, Joshua
AU - Samosir, Jistrani
AU - Amalia, Ridha
AU - Yulianti, Mira
AU - Pitoyo, Ceva Wicaksono
AU - Setiati, Siti
N1 - Publisher Copyright:
© 2023
PY - 2023/6
Y1 - 2023/6
N2 - COVID-19-related acute respiratory distress syndrome (CARDS) has been suggested to differ from the typical ARDS. While distinct phenotypes of ARDS have been identified through latent class analysis (LCA), it is unclear whether such phenotypes exist for CARDS and how they affect clinical outcomes. To address this question, we conducted a systematic review of the current evidence. We searched several, including PubMed, EBSCO Host, and Web of Science, from inception to July 1, 2022. Our exposure and outcome of interest were different CARDS phenotypes identified and their associated outcomes, such as 28-day, 90-day, 180-day mortality, ventilator-free days, and other relevant outcomes. We identified four studies comprising a total of 1776 CARDS patients. Of the four studies, three used LCA to identify subphenotypes (SPs) of CARDS. One study based on longitudinal data identified two SPs, with SP2 associated with worse ventilation and mechanical parameters than SP1. The other two studies based on baseline data also identified two SPs, with SP2 and SP1 were associated with hyperinflammatory and hypoinflammatory CARDS, respectively. The fourth study identified three SPs primarily stratified by comorbidities using multifactorial analysis. All studies identified a subphenotype associated with poorer outcomes, including mortality, ventilator-free days, multiple-organ injury, and pulmonary embolism. Two studies reported differential responses to corticosteroids among the SPs, with improved mortality in the hyperinflammatory and worse in the hypoinflammatory SPs. Overall, our review highlights the importance of phenotyping in understanding CARDS and its impact on disease management and prognostication. However, a consensus approach to phenotyping is necessary to ensure consistency and comparability across studies. We recommend that randomized clinical trials stratified by phenotype should only be initiated after such consensus is reached. Short title: COVID-19 ARDS subphenotypes and outcomes.
AB - COVID-19-related acute respiratory distress syndrome (CARDS) has been suggested to differ from the typical ARDS. While distinct phenotypes of ARDS have been identified through latent class analysis (LCA), it is unclear whether such phenotypes exist for CARDS and how they affect clinical outcomes. To address this question, we conducted a systematic review of the current evidence. We searched several, including PubMed, EBSCO Host, and Web of Science, from inception to July 1, 2022. Our exposure and outcome of interest were different CARDS phenotypes identified and their associated outcomes, such as 28-day, 90-day, 180-day mortality, ventilator-free days, and other relevant outcomes. We identified four studies comprising a total of 1776 CARDS patients. Of the four studies, three used LCA to identify subphenotypes (SPs) of CARDS. One study based on longitudinal data identified two SPs, with SP2 associated with worse ventilation and mechanical parameters than SP1. The other two studies based on baseline data also identified two SPs, with SP2 and SP1 were associated with hyperinflammatory and hypoinflammatory CARDS, respectively. The fourth study identified three SPs primarily stratified by comorbidities using multifactorial analysis. All studies identified a subphenotype associated with poorer outcomes, including mortality, ventilator-free days, multiple-organ injury, and pulmonary embolism. Two studies reported differential responses to corticosteroids among the SPs, with improved mortality in the hyperinflammatory and worse in the hypoinflammatory SPs. Overall, our review highlights the importance of phenotyping in understanding CARDS and its impact on disease management and prognostication. However, a consensus approach to phenotyping is necessary to ensure consistency and comparability across studies. We recommend that randomized clinical trials stratified by phenotype should only be initiated after such consensus is reached. Short title: COVID-19 ARDS subphenotypes and outcomes.
KW - ARDS
KW - Conference presentation: none
KW - COVID-19
KW - COVID-19 ARDS
KW - Outcomes
KW - Phenotyping
UR - http://www.scopus.com/inward/record.url?scp=85162089294&partnerID=8YFLogxK
U2 - 10.1016/j.heliyon.2023.e17276
DO - 10.1016/j.heliyon.2023.e17276
M3 - Article
AN - SCOPUS:85162089294
SN - 2405-8440
VL - 9
JO - Heliyon
JF - Heliyon
IS - 6
M1 - e17276
ER -